Anal. Chem. 2010, 82, 8374–8376
Letters to Analytical Chemistry Viscoelastic Sensing of Conformational Changes in Plasminogen Induced upon Binding of Low Molecular Weight Compounds Erik Nileba¨ck,† Fredrik Westberg,‡ Johanna Deinum,‡ and Sofia Svedhem*,† Department of Applied Physics, Chalmers University of Technology, 412 96 Go¨teborg, Sweden, and Bioscience, AstraZeneca R&D, 431 83 Mo¨lndal, Sweden Plasminogen is a precursor to the fibrinolytic enzyme plasmin and is known to undergo large conformational changes when subjected to low molecular lysine analogues such as tranexamic acid (TA) or ε-amino-n-caproic acid (EACA). Here, we demonstrate how well-controlled surface immobilization of biotinylated plasminogen allows for monitoring of the interaction between TA and EACA with plasminogen. The interaction was studied by the quartz crystal microbalance with dissipation monitoring (QCM-D) technique as well as by surface plasmon resonance (SPR) based sensing. QCM-D measures changes in acoustically coupled mass (by detection of changes in the resonance frequency of the crystal, ∆f) and is sensitive to changes in mass adsorbed on the sensor surface including how liquid medium is associated with this material. Through the dissipation factor (i.e., changes in the energy dissipation of the crystal oscillation, ∆D), QCM-D is also sensitive to the viscoelastic properties of material adsorbed to the sensor surface. Upon binding of TA or EACA, changes in the plasminogen structure were recorded as distinct, although small, ∆D responses which were used to determine affinity constants. By comparing native and truncated plasminogen, we conclude that the observed dissipation shifts were caused by conformational changes in the proteins leading to changes in the viscoelastic properties of the protein layer on the surface. These results demonstrate a novel application of the QCM-D technique, paving the way for a whole new approach to screening of this target for novel lead structures. Proteins and their interactions with other biomolecules (proteins or other) are involved in virtually all physiological processes, which is also why they are the main targets for the vast majority of drug discovery research.1 The function of proteins is closely related to their structure, and their interactions typically induce conformational changes of the protein. A prominent example is the plasmin precursor * To whom correspondence should be addressed. Phone: +46 31 772 3428. Fax: +46 31 772 3134. E-mail:
[email protected]. † Chalmers University of Technology. ‡ AstraZeneca R&D. (1) Arkin, M. R.; Wells, J. A. Nat. Rev. Drug Discov. 2004, 3 (4), 301.
8374
Analytical Chemistry, Vol. 82, No. 20, October 15, 2010
Figure 1. Schematic representation of the conformation change in biotinylated native plasminogen (biotin-Glu-PLG) immobilized to steptavidin (SA) when subjected to a lysine analogue such as tranexamic acid (TA) or ε-amino-n-caproic acid (EACA). A selfassembled monolayer of thiolated polyethylene glycol (PEG) onto gold exposing biotin groups was used to immobilize SA (not to scale).
plasminogen (PLG).2 PLG is transformed into the enzyme plasmin, which degrades fibrin, on clot surfaces by the action of tissue plasminogen activator (t-PA), and by binding to fibrin the affinity of PLG for t-PA becomes enhanced. PLG contains one protease domain and five different kringle domains, each containing a lysine binding site involved in fibrin binding. When certain low molecular weight lysine analogues bind to one of the five lysine binding sites, PLG undergoes a large, reversible, conformational change from a prolate ellipsoid to a structure better described as a random coil (Figure 1).2 Two examples of such compounds are tranexamic acid (TA) and ε-amino-n-caproic acid (EACA), which have been shown to bind to the low affinity lysine binding sites of PLG and thereby prevent binding of PLG to fibrin, thus inhibiting fibrinolysis. Both compounds are therefore used clinically to stop excessive bleeding; TA is known to be more potent than EACA.3 The conformational changes that occur in PLG upon occupation of a low affinity lysine binding site (2) Mangel, W. F.; Lin, B.; Ramakrishnan, V. Science 1990, 248 (4951), 69. (3) Casati, V.; Guzzon, D.; Oppizzi, M.; Cossolini, M.; Torri, G.; Calori, G. Ann. Thorac. Surg. 1999, 68 (6), 2252. 10.1021/ac1016419 2010 American Chemical Society Published on Web 09/20/2010
Table 1. Quantification of the Immobilized Amounts of Protein in QCM-D and SPR Experimentsa protein
∆f7/7 (Hz)
∆D7 (10-6)
∆SPR (RU)
massQCM-D (ng/cm2)
hydrationb (%)
SA Glu-PLG Lys-PLG
-22.0 ± 1.3 -38.3 ± 5.5 -48.6 ± 16
0.03 ± 0.12 2.1 ± 0.33 2.4 ± 0.86
1835 ± 42 2521 ± 254 2082 ± 366
432 ± 41 1205 ± 109 1511 ± 393
58 78 86
a Also showing the corresponding dissipation and calculated degree of hydration for the immobilized protein layers. Adsorption values from QCM-D are shown as normalized frequency shifts for the 7th overtone that was chosen as the most reproducible and stable overtone. b The degree of hydration was calculated by comparing the acoustic mass calculated from ∆f using the Voigt model10(QCM-D) and by assuming 1 kRU ) 100 ng/cm2 for SPR.10 The value obtained for SA was similar to previous studies.7,10
have been investigated by several techniques; small-angle neutron scattering,2,4,5 sedimentation,6 X-ray crystallography,5 size-exclusion high-performance liquid chromatography,5 dynamic light scattering,5 and circular dichroism.2 These studies show that the radius of gyration increases, from 39 to 56 Å for native PLG (also referred to as Glu-PLG, Figure 1)2,4 and from 51 to 52 Å for the truncated version of the protein (Lys-PLG) where the first 77 amino acids have been removed by plasmin.4 The difference in radius of gyration between Lys-PLG and Glu-PLG suggests that the large conformational change already has occurred with the removal of the first 77 amino acids in Lys-PLG while retaining its lysine binding sites.4 In this study, surface immobilized PLG was subjected to EACA and TA and the interactions were monitored by the quartz crystal microbalance with dissipation (QCM-D) technique. QCM-D7 is based on sensing of mass which is acoustically coupled to the sensor surface, and it is known to monitor structural rearrangements in thin films, e.g., induced during protein adsorption or upon cross-linking of proteins8 and layer-by-layer assemblies.9 Thus, we hypothesized that changes in the conformational state of immobilized Glu-PLG would promote a change in viscoelastic properties different from Lys-PLG, detectable by the dissipation factor in QCM-D. Complementary data were obtained by an optical technique based on surface plasmon resonance (SPR). EXPERIMENTAL SECTION Biotinylated PLG was immobilized to gold coated sensor surfaces via the protein streptavidin (SA), which is a common linker between a biotinylated sensor surface and a biotinylated biomolecule. Biotinylated sensor surfaces were prepared through self-assembly of a functionalized monolayer on the gold coated surfaces before mounting in the instrument, after which SA followed by biotinylated Lys-PLG or Glu-PLG were injected over the sensor surfaces (Table 1). The plasminogen immobilization was not allowed to reach complete saturation, which could potentially sterically affect the unfolding of the protein. After the PLG had been immobilized, the protein layers were subjected to increasing concentrations of TA and EACA (ranging from 0.005 to 5 mM for TA and from 0.005 to 10 mM for EACA). For experimental details, see the Supporting Information. (4) Ramakrishnan, V.; Patthy, L.; Mangel, W. F. Biochemistry 1991, 30 (16), 3963. (5) Marshall, J. M.; Brown, A. J.; Ponting, C. P. Biochemistry 1994, 33 (12), 3599. (6) Markus, G.; Priore, R. L.; Wissler, F. C. J. Biol. Chem. 1979, 254 (4), 1211. (7) Hook, F.; Ray, A.; Norden, B.; Kasemo, B. Langmuir 2001, 17 (26), 8305. (8) Hook, F.; Kasemo, B.; Nylander, T.; Fant, C.; Scott, K.; Elwing, H. Anal. Chem. 2001, 73 (24), 5796. (9) Vogt, B. D.; Lin, E. K.; Wu, W. L.; White, C. C. J. Phys. Chem. B 2004, 108 (34), 12685. (10) Larsson, C.; Rodahl, M.; Hook, F. Anal. Chem. 2003, 75 (19), 5080.
RESULTS AND DISCUSSION With comparison of the QCM-D and SPR data, the water content of the immobilized protein layers was determined (Table 1). The degree of hydration of Lys-PLG was found to be higher than for GluPLG, supporting that Lys-PLG was at the beginning of the experiment in a different structural state than Glu-PLG. When subjecting immobilized Glu- and Lys-PLG to increasing concentrations of TA and EACA, increasing and reversible dissipation shifts were observed (Figure 2, blue and red curves). The maximum ∆D responses were always higher for Glu-PLG (0.6 × 10-6) compared to Lys-PLG (0.3 × 10-6), and the more potent PLG activator TA had an effect on the dissipation response at lower concentrations than the less potent EACA. A simple 1:1 model was used for the fitting since the conformation change is known to occur after binding to only one weak lysine binding site on plasminogen, even though there are five available binding sites in total. From the titration curves shown in Figure 2, apparent Kd values for the binding of TA were calculated at 125 µM (Lys-PLG) and 300 µM (Glu-PLG). The corresponding Kd values for EACA were 1400 µM (Lys-PLG) and 1700 µM (Glu-PLG). Similar Kd values have been reported for these interactions in the literature with other techniques,5,6 which supports that the observed dissipation shifts result from the interaction of PLG with TA and EACA. Other than the maximum dissipation shift, Glu-PLG differs from Lys-PLG in that small, reversible frequency shifts (∆f ≈ -1 Hz) were observed for Glu-PLG at ligand exposure at the higher concentrations. For Lys-PLG, no frequency shifts were observed. Additionally, in SPR measurements, the responses upon TA/EACA binding were close to the detection limit for all concentrations and for both PLG variants. The QCM-D responses obtained for the interaction between the two PLG variants and the low molecular weight specific ligands TA and EACA cannot be explained by mass uptake per se. In SPR measurements, the optical responses were
